[PATCH] Fix darwin bootstrap, also some testcases too

Stan Shebs shebs@apple.com
Wed Aug 18 00:10:00 GMT 2004


Andrew Pinski wrote:

>
> On Aug 17, 2004, at 4:19 PM, Stan Shebs wrote:
>
>> Andrew Pinski wrote:
>>
>>> As I said before some of the patterns were not needed on the FSF 
>>> mainline
>>> (even some of them are really not needed on the apple-ppc-branch 
>>> also but
>>> that is for another day).
>>>
>>> Note when save world gets committed to the mainline, the save world 
>>> part
>>> of this patch can now be committed again but this time using the
>>> constraint, save_world_operation instead.
>>>
>>> OK? Andreas Tobler bootstrapped and tested on powerpc-apple-darwin.
>>
>>
>> Is this known to work for non-darwin configs? Newlib building falls 
>> over with
>> an ICE for x-powerpc-eabi even with the darwin stuff backed out, so 
>> I'm not
>> confident of anything yet.
>
>
> Is the ICE in dojump.c, if so this is a different issue and a known 
> problem, see PR 17066.
>
Yes, that's the one.

> Again this patch just removes the patterns which should have never 
> made it in
> because they don't do anything for the FSF mainline.
>
> Note also the patch you applied to the apple-ppc-branch also have the 
> same issues
> but don't show up because the string instructions are disabled even 
> with -Os.

In any case, the patch looks fine to me, but I'd like somebody else to 
rule on it.

Stan



More information about the Gcc-patches mailing list